I was about to rant about this, about how 0.97 is virtually the same as 1, but I note that the unit used is a point on the SF-36-PFS scale.also, the attempt to find predictors of improvement at 12 months, came to nothing, with odds ratios of 0.97 and 0.96 baseline pain and SF 36 respectively.
So, for example, for every point higher in the SF-36 -PFS at baseline, there was a 3% reduction in the odds of achieving a clinically important difference at 6 months. If someone has a 10 point higher score at baseline than another patient, that equates to a substantially lower chance of achieving a clinically important improvement at 6 months. Of course, it's not very surprising - someone with a low score has a better chance of improving a lot than someone with a higher score, reversion to the mean and all that.
It is interesting that they didn't find any difference in improvement related to age or gender.
They must have been really disappointed by those results. Baseline levels of anxiety and depression weren't found to make a difference to the chance of a clinically important improvement.Seven other logistic models were run, which included, as individual covariates; months since onset, HADS anxiety, HADS depression, SCAS, CFQ, school attendance and pain at baseline (table 5, also adjusted by age, gender and baseline SF-36-PFS). However, none of these models showed any evidence of an effect of the individual covari- ates on the odds of reaching the MCID at 6 months, and, therefore, we believe that the effect of baseline SF-36-PFS was not confounded by these variables.